A novel DEM enhancement methodology to improve physics-based susceptibility modeling of rainfall-induced landslide along anthropogenically modified slopes

Rainfall-induced landslide is a pervasive hazard in steep, high-elevation terrains, where localized anthropogenic modifications implemented without adequate drainage systems increase pore water pressure, significantly exacerbating slope instability. The high expense and limited availability of fine-resolution data often hinders regional slope stability models from accounting for such fine-scale terrain details, leading to reduced predictive accuracy. This study presents a novel methodology to enhance digital elevation models (DEMs) for capturing fine-scale terrain features, focusing on terraced slopes. Using high-resolution satellite imagery, the approach integrates advanced image processing, particularly the Canny edge detection algorithm for precise terracing pattern recognition, combined with field measurements and geospatial analyses. Validation against LiDAR-derived DEMs yielded an R 2 value of 0.9876, with extracted slopes exhibiting a mean relative error of 0.28, substantially outperforming conventional resampling methods. The framework integrating DEM enhancement and physics-based model was applied to two landslide case studies in the Western Ghats, India: 2021 Plappally and 2019 Kavalappara landslides. Results demonstrated this framework effectively delineates failure zones and captured critical pore pressure amplification due to terracing, aligning well with field observations and existing coupled flow-deformation analyses. This scalable methodology significantly enhances slope stability analyses and landslide hazard assessments in regions with anthropogenic slope modifications.